Stephen Jay Kline, Conceptual Foundations for Multidisciplinary Thinking, Stanford, CA: Stanford University Press, 1995.
© Systems Research, 13(4): 491-503 (1996)
This unpretentious book — plainly written, packaged with a straight-forward descriptive title and published by Stanford University Press with little fanfare — should be required reading for every college student — and their professors. It is that important. Unfortunately, such an eventuality is extremely unlikely, not least because this book is also deeply subversive to “business as usual” in the academic world.
Stephen Jay Kline, an emeritus professor at Stanford, is one of the four founders of Stanford’s renowned multidisciplinary undergraduate program — now called Science, Technology and Society — as well as a long-time faculty member in Stanford’s Department of Mechanical Engineering. For over a quarter of a century, he and his pioneering colleagues have been seeking to provide an antidote to the tendency toward ever-greater specialization and disciplinary Balkanization in academia, especially in the sciences. Kline has invested heavily in cross-disciplinary scholarship and in developing a multidisciplinary framework, which he has field-tested with several generations of students. His book represents a distillation of that “on-line” experience.
Kline’s thesis, in a nutshell, is that neither reductionist approaches nor holistic/systemic approaches (he uses the less familiar term “synoptic”) are sufficient, although each can contribute to our understanding in important ways. But more important, Kline shows in detail exactly why a multidisciplinary framework is ultimately essential, not just an “elective”. In effect, he brings a rigorous analysis to the “case” for the pragmatic, common-sense perspective that many of us have been systematically educated to ignore. This in itself is a major contribution. But Kline does much more. The dust jacket for the book displays a compass rose as a design motif, and this is an appropriate metaphor for the fact that Kline also provides his readers with practical aids to navigation — specific tools and illustrative examples. The book is as much a repair manual (to shift metaphors) as it is a diagnostic exercise, not to mention being also a short history of how the academic world got into its present state. In short, this book is a one-of-a-kind.
The key elements of Kline’s argument can be briefly summarized. One fundamental point is that the basic structure of the phenomenal world is multi-levelled, with “emergent” properties and “degrees of freedom” that cannot be adequately described, predicted or explained in terms of lower-level phenomena. Moreover, the various levels of organization are linked to one another by “interfaces of mutual constraint” (as he terms it) — both upward and downward. Kline shows how “lower” levels may constrain but cannot determine the properties of “higher” levels. This much covers relatively familiar ground. Similar arguments have been made by (among others) Michael Polanyi 1968; Roger Sperry 1969, 1991; Paul Weiss 1971; Perry Anderson 1972; Howard Pattee 1973; William Wimsatt 1974; William Bechtel 1986; as well as by the founding fathers of the systems sciences (e.g, Ludwig von Bertalanffy 1968; Kenneth Boulding 1956; Anatol Rapoport 1968; and James G. Miller 1995). However, Kline argues the case in greater depth and more compellingly than I have seen anywhere else. As he points out, a full-court press has become necessary because the majority of working scientists, and their students, still have not incorporated this elemental aspect of the natural world into their thinking. Witness the claims, as recently as the beginning of 1996, that theoretical physics may be on the verge of producing “a theory of everything.” Kline is particularly caustic towards what he characterizes as “disciplinary overclaims.”
A second key point, equally crucial to Kline’s argument, is the fact that our attempts to comprehend the phenomenal world are severely constrained by (a) the inherent complexity of the subject-matter and (b) the severe limitations, and the quirks, of the human mind (the evidence for which he describes in some detail). Among other things, he reminds us that the humans are very good at associative learning (as the Behaviorist psychologists have documented to death) but are rather weak, by and large, at deductive reasoning (Sherlock Holmes notwithstanding). Scientists often lose sight of the fact that they are only creating imperfect mental representations (“schemata”) of the real world, and forget — if they ever knew — Korzibski’s dictum that “the map is not the territory.” Kline shows in embarrassing detail (at least to some of the perpetrators) how the tendency to discount our own intellectual limitations has been responsible for a lot of muddled thinking and misplaced hubris in the sciences.
One of the more famous examples of this is used by Kline in a highly unusual way. He offers a formal $1,000 bet at odds of 10:1 (the rules are very specific and the money is being held in an escrowed bank account) to any physicist who can develop a complete, logically correct statement of the Second Law of Thermodynamics from statistical mechanics. The significance of this bet derives from the fact that the Second Law has different properties at the “classical” macroscopic level covered by Rudolph Clausius’s formulation and in the “statistical analogues” that were developed independently by Ludwig Boltzmann and Willard Gibbs to describe the operation of the Second Law at the “microscopic” level of individual atoms. One glaring aspect of this disjunction between the different levels of analysis is the century-old “irreversibility paradox.” At the micro-level, the formal equations that physicists have developed over the years allow for reversibility in a thermodynamic process (under carefully specified model conditions, that is). Yet, at the macro-level, such processes are always irreversible, both in theory and in our empirical observations. Which version of the Second Law is correct? Some theorists — even some of our leading physicists — have maintained that the paradox is real, a perplexing and deeply mysterious property of nature itself. Kline argues that, to the contrary, the paradox is an artifact of faulty reasoning, and of conflating our imperfect mental representations with the more logically-consistent properties of the phenomena we study.
A third major point developed by Kline is that cybernetic, information-using, feedback-controlled processes are a fundamentally-important aspect of the living world in general and especially of human behavior and human-designed systems. Moreover, cybernetic processes cannot be “reduced” to a lower level or even fully explained within a deterministic paradigm. Indeed, cybernetic processes cannot be “explained” at any one level, or by any one discipline.
To cite an example — not used by Kline but one that I am sure he would endorse: Consider the problem of explaining — fully — the properties and the behavior of the fire in my fireplace. One would have to begin with the principles of atomic physics and go on to (among others) quantum mechanics, solar physics, thermodynamics, molecular biology, biochemistry, plant morphology and physiology, ecology, climatology, geology and materials science, industrial engineering, architecture, cultural anthropology and the history of human technology, as well as evolved construction methods and local building codes.
But that is only the beginning. One must also account for the influence of human skills and human actions, including the multiple steps associated with cutting, distributing and purchasing the firewood and obtaining matches (economic activity), and, finally, the actions associated with building the fire, opening the flue, starting the fire and tending it, all of which involve complex cybernetic interactions and feedbacks within and between levels. (For instance, I will stoke the fire and add wood depending on how it “performs”.) And we have left out of this catalog one of the most important causes of all, namely, human purposes — my physiological need for warmth, my aesthetic preferences, and the instrumental goals that derive from these motivations. Moreover, despite our vast superstructure of scientific knowledge, nobody (not even I) can predict when I will build my next fire. It might depend on the vagaries of the weather and on when some friends next come to visit.
Some of the many insights, hypotheses, propositions and dicta that Kline proposes, or compiles, in this volume include the following:
- A “system” can represent whatever aspects of the phenomenal world that we want it to, and much confusion and/or myopia in science can be traced to a failure to specify properly where the theoretical system begins and ends, and why. Any truth claim about the phenomenal world should specify which part is being referred to, and which parts are being excluded.
- Neither reductionist nor holistic (synoptic) approaches to the analysis of complex phenomena can ever be sufficient, for a number of reasons. Indeed, Kline devotes entire chapters to such issues as the theoretical implications of hierarchical organization and mutual constraints, the need for dimensional homogeneity in analyzing hierarchical systems (Kline cites physicist Niels Bohr’s frequently-neglected “correspondence principle”), and the implications of control information and feedback in living systems and their techno-economic extensions.
- The awesome, irreducible complexity of human systems argues for the absolute necessity of making data-based, incremental changes rather than using top-down grand theories or vacuous mathematical formalizations. Kline cites some examples of how the failure to heed this dictum has led to disastrous consequences.
- The individual disciplines need to talk to one another on a sustained basis. For instance, Kline counted six very different models of innovation processes in each of six different academic disciplines (and he left out a seventh, biology). He is particularly critical of economics, where much of the theoretical work involves what biologist John Maynard Smith, referring to some of the mathematical modellers in his own discipline, called “data free science.” Kline’s own model of economic innovation, the so-called “Chain-Linked Model,” provides an illustration of how a multidisciplinary approach can lead to a more sophisticated, empirically-grounded theoretical framework.
Kline’s book also provides a number of practical “tools”. In addition to the model of innovation processes noted above, Kline also develops an “Index of Complexity” that breaks new ground on this much-debated issue. He proposes to add a functional criterion — the number of feedback loops in a system — to other, more conventional physical measures. (His index is already in use among some of Kline’s colleagues.) Kline also provides some sorely needed criteria by which scholars from outside a given discipline can evaluate the appropriateness of an insider’s paradigms and theoretical assumptions — potentially a powerful instrument for breaking down the insularity (and, sometimes, the naivete) of various specialized fiefdoms. Kline also devotes two chapters to analyzing and classifying the problems that are endemic to the fragmented nature of our scientific enterprise. Finally, in an appendix section, Kline proposes a relatively small institutional change in the faculty structure of our universities that could materially improve substantive communications between the disciplines and, ultimately, the quality of undergraduate education. It is a proposal that deserves to be considered seriously.
There are very few things to criticize in this profoundly important volume. Some repetitiousness and idiosyncratic terminology could easily be corrected. The occasional lapses in proof-reading and presentation are mostly minor, with one significant exception that is likely to cause some confusion. In two different places Kline uses terminology in his explications regarding the dimensionality of physical systems which are inconsistent with his diagrams (Figures 4.1 and 5.1). In the text, he chooses to use the term “B” to denote length, while “L” is assigned to feedback loops. But, in the diagram, length is labelled “L”. Adding to the confusion, later on in the discussion length is designated “l”, while the term “L” is used with reference to hierarchical levels and feedback is designated “f” and “F”. In future editions, which this volume amply deserves, such problems can easily be rectified.
A more serious criticism derives from the fact that, despite his impressive command of a broad range of disciplines, Kline retains a physical science bias that colors his presentation, if not his world view. The root of the problem, I believe, lies in the fact that Kline made an insufficient investment over the years in the biological sciences. The results are reflected in many ways, some of them subtle and others quite blatant. Thus, in his discussion of the dimensionality of physical systems, he simply adds feedback loops to the framework without acknowledging the fact that this involves a fundamental shift of system types. Feedback controlled systems are purposive; a feedback loop is always connected to purposes/goals/end-states (which Kline spends little time discussing). Indeed, in the classic example of a feedback system (dating back to the cybernetics pioneer, Norbert Wiener) — namely, a room with a thermostat connected to a furnace — the purposive basis of the system and its behavior becomes apparent only if the human designer, builder and controller (the person who adjusts the thermostat settings and pays the heating bills) are added to the system (as Kline himself admonishes us to do).
One consequence of this bias is that Kline treats the causal role of biological purposiveness only obliquely. He refers frequently to the mathematical concept of “degrees of freedom”, and he references the role of “information” and “choices” in human-designed systems, but there is no systematic treatment of the basis of human choices. Indeed, he denounces what has been called “vulgar sociobiology” — the extremist argument for biological determinism in social behavior — apparently without appreciating the more balanced literature on the biological bases of behavior which stresses the complexity of organism-environment interactions. (One notable exception is Kline’s reference to the classic work of psychologist Harry Harlow on “love” in infant monkeys). This research domain was reviewed in some detail in my own volume on The Synergism Hypothesis (1983), but there are many more recent studies of the subject in disciplines ranging from ethology to psychobiology, behavior genetics and evolutionary psychology.
Another example of how Kline’s presentation was shaped by a physical science perspective is his treatment of the relationships between hierarchical “levels” in complex systems. He embraces physical chemist Michael Polanyi’s (1968) concept of “mutual constraints” (he calls it “Polanyi’s Principle”), and certainly these are very important. (Consider, for instance, how our actions are constrained by gravity, at least on earth.) Yet he shies away from the additional point that the different levels of organization also “interact” in a great many ways, with many complex feedback relationships as well. One illustration is the fireplace example used above.
Finally, Kline’s perspective also influenced his treatment of time. In biology, time and the accumulated contingencies (and legacies) of history are of fundamental importance. Time is not simply another dimension, like length and width. Time leads to changes not only in the variables; it also changes the parameters. The explanatory edifice that is required to account for the paradigmatic fire in my fireplace also involves an historical element, the influence of a vast accumulation of relevant past events and, equally important, a proximate, historically unique configuration of contextual circumstances. If history were “reducible” to a universal generalization (much less a dimension) that is any more precise than “natural selection” or “functional synergies”, most likely by now somebody would have thought of it. (Many “laws” of history have been proposed, but none so far have been adopted.) In any case, history is an inescapable aspect of the biological and human sciences.
However, it should be emphasized that this criticism too is remediable in future editions, albeit at greater cost to both the author and the publisher. To my mind, it would perfect what is already a major contribution to the overarching goal that, as Kline argues convincingly, is ultimately an imperative for long-term scientific progress — a multidisciplinary perspective.